Data, Demand, and Demographics: A Symposium on Housing Finance
Co-presented by the Urban Institute and CoreLogic
November 2, 2016
Welcome Laurie Goodman – Director, Housing Finance Policy Center, Urban Institute Faith Schwartz – Principal, Housing Finance Strategies, representing CoreLogic
Join The Conversation #HousingData 2
Urban Challenges and Opportunities Policies of Permitting, Preserving, and Advancing our Cities
Ed Glaeser – Fred and Eleanor Glimp Professor of Economics, Harvard University
Join The Conversation #HousingData 3
New (and Old) Directions in Housing Market Research Edward L. Glaeser Harvard University
4
Is the Bubble Back? (S+P, C-S, Corelogic)
5
Three Themes • The Rise of Sub-City Data and Neighborhood Measurement Tools (with Cesar Hidalgo, Nikhil Naik, Scott Kominers) • The Growing Academic Consensus on Real Estate Bubbles (with Charles Nathanson) • The Underappreciated Power of Supply (with Joe Gyourko, also Chinese material joint with Andrei Shleifer, Yueran Ma, Wei Huang)
6
FHFA Zip Code Data From Beogen, Derner And Larson
7
Hipsman (2015) Using Zillow Sub-city
8
The Promise of Google Street View + Computer Vision (with C. Hidalgo, N. Naik and S. Kominers)
• Google Street View has covered more than 3,000 cities from 100 countries across the world • High resolution imagery at street-level: amenable for analysis by both humans and computers • Time series: 2008 - Present • Data from India and China should be available soon (from various providers), already available for Brazil, Indonesia etc.
9
The Promise of Google Street View + Computer Vision
• Availability of Street View paralleled by impressive gains in computer vision technology, fuelled by deep learning. • Opportunity to develop automated surveys of the built environment at unprecedented resolution and scale
10
Streetscore (Cesar Hidalgo): How safe does this place look to humans?
1.8/10
9.2/10
Goal: Train a computer to assign a score to a street view image for “perceived safety” from image pixels Can be extended to “perceived” wealth, liveliness, cleanliness etc.
11
Train A Computer Vision Model to Predict Streetscore (Perceived Safety FROM Naik) Training Examples
8/10
Computer Vision
Predicted Streetscore
5/10
3/10
Image Features Derived from Pixels
6.4/10
12
Computing Urban Change
• Urban Change Coefficient (UCC) : Change in Streetscore of images of the same location between 2007 and 2014
1.8/10
9.2/10
UCC = +7.4 (positive change) Naik et.al., NBER Working Paper, 2015
13
Urban Change Coefficient Significant decay
14
Urban Growth in New York City 2007 - 2014
15
Which demographic factors precede physical urban change? With Hidalgo, Kominers, Naik and Raskar (2015)
• 5 cities – 2,514 census tracts • Socioeconomic data from Census • Multivariate spatial regressions
16
17
18
High Frequency Momentum, Low Frequency Mean Reversion
19
Beliefs follow Past Price Growth
20
Irrationality Just Fits the Data Better
21
Figure 13: Mean Reversion across Zip Codes (Mean Residuals from Hedonic Regression) .
Log Value Change 1928-1935 0
10040
10012
10010
-.2 10009
10039 10023 10032 10011 10026 10017 10003 10030 10036 10021 10029 10025 10016 10027 10024 1000210035 10128 10028 10033 10019 10007 10037 10075 10031 10013 10022 10014 10001 10065
10006
10018
10034
-.4
10004
10005
10038
10020
-.6 .1
.2
.3 Log Value Change 1921-1928
.4
.5
Source: Data from Nicholas and Scherbina (2011)
22
.25
Change in Housing Prices, 2001-2006 vs. 2006-2011
Change in FHFA Price, 2006-2011 -.75 -.5 -.25 0
Houston New York DC Detroit Phoenix
-1
Las Vegas
0
.2
.4 .6 Change in FHFA Price, 2001-2006
.8
23
Multi Family Permits Single Family Permits
24
Nov 2014
Sep 2013
Jul 2012
May 2011
Mar 2010
Jan 2009
Nov 2007
Sep 2006
Jul 2005
May 2004
Mar 2003
Jan 2002
Nov 2000
Sep 1999
Jul 1998
May 1997
Mar 1996
Jan 1995
Nov 1993
Sep 1992
Jul 1991
May 1990
Mar 1989
Jan 1988
Nov 1986
Sep 1985
Jul 1984
May 1983
Mar 1982
Jan 1981
Nov 1979
Sep 1978
Jul 1977
May 1976
Mar 1975
Jan 1974
Nov 1972
Sep 1971
Jul 1970
May 1969
Mar 1968
Jan 1967
Nov 1965
Sep 1964
Jul 1963
May 1962
Mar 1961
Jan 1960
Single Family and Multi-Family Permits Over Time
2,000
1,800
1,600
1,400
1,200
1,000
800
600
400
200
0
25
26
27
28
29
30
31
The Great Chinese Housing Boom
People lining up outside a residential project before sales start. Hefei, Anhui Province.
32
33
34
35
36
37
Data, Demand, and Demographics: A Symposium on Housing Finance
Co-presented by the Urban Institute and CoreLogic
November 2, 2016
Housing and Economic Outlook 2016 and Beyond
Frank Nothaft – Senior Vice President and Chief Economist, CoreLogic
Join The Conversation #HousingData 39
Housing and Economic Update: 2016 and Beyond Data, Demand, and Demographics: A Symposium on Housing Finance Frank Nothaft, CoreLogic SVP & Chief Economist November 2, 2016
40 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
The ‘New Normal’ 1. Low mortgage rates are the norm 2. Household composition increasingly diverse 3. Sales rise but turnover remains below ‘average’ 4. Originations: Purchase & HELOC up, Refi down 5. Loan performance excellent (new credit ‘lower risk’) 41 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
When Will the Fed Raise Target Rates? (Soon) Federal Funds Target (interest rate, in percent) 4.0 3.5
Minutes Sep 21, 2016 Sep 21, 2016 Median
3.0 2.5 2.0 1.5 1.0 0.5 0.0 Sept. 21, 2016 Median
2016
2017
2018
0.625
1.125
1.875
2019 2.625
Longer Run 2.875
Source: Federal Open Market Committee Meeting on September 21, 2016. In the plot each circle indicates the value (rounded to 42 the nearest 1/8 percentage point) of an individual FOMC participant’s judgment of the appropriate level of the target federal funds rate at the end of the specified calendar year or over the longer run. ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
Low Mortgage Rates Are the New Norm Interest Rate on 30-Year Fixed-Rate Mortgages (percent) 7%
Forecast
6%
Dec. 2017:
5%
4%
4.2%
Great Recession
3% 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Source: Freddie Mac Primary Mortgage Market Survey®, IHS Global Insight October 2016 projection.
43 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
Large Demographic Tailwind Has Arrived Population in 2015 (Millions)
Largest Age Cohort
4.8 Average Age Firsttime Homebuyer
4.6 4.4
Average Age Repeat Buyer
4.2 4 3.8 3.6 3.4
18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40
Age in 2015 Source: U.S. Census Bureau, Population as of July 1, 2015
44
©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
Household Composition More Diverse Three-fourths of New Households Will Be Minority-Headed 2015 Share of All Households
2015-25 Share of Household Growth Asian & Other 18%
Hispanic Asian & Other 7% 13% Black 12%
White 68%
117 Million Households in 2015
Hispanic 40%
White 24%
Black 18%
12 Million Increase by 2025
Source: Census Bureau Housing Vacancy Survey (2015 household count), Harvard University Joint Center for Housing Studies (Baseline Household Projections for the Next Decade and Beyond, Working Paper w14-1)
45
©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
Sales Rise but Home “Turnover” Remains Low Does ‘New Normal’ Have Lower Turnover? Home Sales as a Percent of Housing Stock 8%
6%
2000-2003 average = 5.6% 4%
2%
0% 2000
2002
2004
2006
2008
2010
2012
2014
2016
2018
Source: CoreLogic REAS MarketTrends through June 2016, Census Bureau HVS, Forecast averages 46 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential. projections of FNMA, FHLMC, Zelman and Associates, MBA, NAR and NAHB.
46
Americans Are Keeping Their Homes Longer Number of Years A Home Is Owned (Median) 14 12
Owner Occupants
10
Home Sellers
8 6 4 2 1985
1990
1995
2000
2005
2010
2015
Source: American Housing Survey for the United States, various years (difference between survey year and median year owner-occupant moved into unit), CoreLogic public records for 47 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential. United States (length of time between recorded sales on same home).
Low For-Sale Inventory: Part of a ‘New Normal’? Homes-For-Sale Inventory as a Percent of Households 4%
3%
2%
1%
0% 1983 1986 1989 1992 1995 1998 2001 2004 2007 2010 2013 2016 Sources: National Association of Realtors, U.S. Census Bureau (New Residential Sales and Housing Vacancy Survey). Note: Existing home inventory excludes Condo & Co-op Inventory before 1999.
48
©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
U.S. Home Prices: 5% Below 2006 Peak Projected to Return to Peak by Early 2018 CoreLogic Home Price Index (January 2000 = 100) 220
-- Forecast --
200
5% 180 160
43%
140 120 100 2000
2002
2004
2006
2008
Source: CoreLogic Home Price Index (November 1, 2016 release)
2010
2012
2014
2016
2018
49 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
Price Growth Faster For Lower-Priced Houses Cumulative Price Growth Through July 2016 (percent) Since July 2015 8%
Since March 2011 60%
6%
45%
4%
30%
2%
15%
0%
0% More Than 25% Below Median
25% or Less Below Median
Price Growth Since:
Up to 25% Above More Than 25% Median Above Median
One Year Ago
Source: CoreLogic HPI, Single-family Detached (November 1, 2016 release); March 2011 is “Post-Great Recession” price trough.
Price Trough
50 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
Originations: More Purchase, Less Refi in 2017 Single-family Mortgage Originations (Billions of dollars) $2,400 Est. --Forecast--
$2,000 $1,600
Refinance
$1,200 $800
Purchase
$400 $0 2009
2010
2011
2012
2013
2014
2015
2016
2017
Source: Originations are an average of the latest projections released by Mortgage Bankers Association, Fannie Mae, Freddie Mac and Zelman & Associates. Fannie Mae as of October 2016. Zelman, Freddie Mac and MBA forecast as of September 2016. 51 2009-2014 are benchmarked to HMDA. Numbers do not include HELOCs. ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
HELOC Volume Up in 2016 Approved HELOCs (Billions of Dollars) $400 $350 $300 $250 $200 $150 $100 $50 $0 2000
2004
2008
2012
2016 (Through August, Annualized)
Source: CoreLogic public records, second-lien HELOCs placed more than 60 days after first lien.
52
©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
Mortgage Credit Risk Along Six Dimensions First-Lien Purchase Money Originations 300
Credit Score Less Than 640
200 Low & No Doc Share
LTV Share 95 And Above
100
0
DTI Share 43 And Above Condo Co-op Share Source: CoreLogic Loan Servicing Database
Benchmark (2001 and 2002 Originations) Non-Owner Occupancy Share
Current (2016:Q2) 53
Source: CoreLogic
©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
Excellent Loan Performance: Part of a ‘New Normal’? Serious Delinquency Rate by Origination Cohort 1999-2003
2004-2008
2009-2014
Source: CoreLogic: March 2016 54 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
The ‘New Normal’ 1. Low mortgage rates: Below 5% next two years 2. Household composition increasingly diverse 3. Sales rise but turnover remains below ‘average’ 4. Originations: Purchase & HELOC up, Refi down 5. Loan performance excellent (new credit ‘lower risk’) 55 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
Where to find more information Look for regular updates to our housing forecast, commentary and data at http://www.corelogic.com/blog @CoreLogicEcon @DrFrankNothaft
The views, opinions, forecasts and estimates herein are those of the CoreLogic Office of the Chief Economist, are subject to change without notice and do not necessarily reflect the position of CoreLogic or its management. The Office of the Chief Economist makes every effort to provide accurate and reliable information, however, it does not guarantee accuracy, completeness, timeliness or suitability for any particular purpose. CORELOGIC and the CoreLogic logo are trademarks of CoreLogic, Inc. and/or its subsidiaries. 56 ©2016 CoreLogic, Inc. All rights reserved. Proprietary and Confidential.
Data, Demand, and Demographics: A Symposium on Housing Finance
Co-presented by the Urban Institute and CoreLogic
November 2, 2016
Panel One
Integrated Services and Inclusionary Housing for Changing Demographics Ellen Seidman – Senior Fellow, Urban Institute (Moderator) Nela Richards – Chief Economist, Redfin Rolf Pendall – Director, Metropolitan Housing & Communities, Urban Institute Jim Carr – Visiting Fellow, Roosevelt Institute Susanne Slater – President and CEO, Habitat for Humanity of Washington, D.C. Judi Kende* – Vice President, Enterprise Community Partners
Join The Conversation #HousingData 58
Integrated Services and Inclusionary Housing for Changing Demographics: Can we build our way out of this? Urban Institute/Core Logic Demand, Data and Demographics Symposium November 2, 2016 Nela Richardson, PhD Chief Economist, Redfin Corp.
59
Five key research findings 1. The housing market is chronically undersupplied (and the rent is too darn high!) 2. Economic mobility depends on geography 3. Economically integrated neighborhoods are rare 4. Land-use regulation affects economic inequality 5. Zoning threatens U.S. productivity and economic growth
60
61
Integrated cities are rare City
Balanced Mix Area
High-end Area
Affordable Area
Boston
51%
35%
15%
Seattle
31%
10%
59%
Washington, DC
30%
25%
45%
San Jose
24%
53%
24%
Denver
24%
7%
69%
San Diego
20%
40%
40%
Los Angeles
19%
74%
7%
Chicago
17%
5%
79%
Austin
16%
11%
73%
Phoenix
13%
11%
76%
Houston
12%
16%
72%
Philadelphia
11%
6%
82%
Baltimore
11%
3%
86%
San Francisco
10%
88%
2%
San Antonio
8%
5%
88%
Memphis
8%
4%
88%
Jacksonville
7%
3%
90%
Detroit
7%
1%
92%
Indianapolis
6%
2%
92%
Columbus
4%
1%
95%
62
Job accessibility in Chicago
63
Families want access to highly ranked schools
64
…and walkable communities
65
Walkable communities are highly valued
66
67
Families are moving farther from the city center
68
69
Thank you! Redfin Research https://www.redfin.com/blog
[email protected] @NelaRichardson
70
Demographics, homeownership, and home equity Trends and policies for high- and low-cost states Rolf Pendall, Ph.D. Co-Director, Metropolitan Housing & Communities Policy Center November 2, 2016
71
72
Homeownership falling, faster in high-cost states Per capita homeownership rates, observed and projected to 2040 75%
50%
25%
0% CA
TX
White NH
CA
TX
Black NH 2010
2020
CA
TX
Other NH 2030
CA
TX
Hispanic
2040
Source: Karen Smith et al., Urban Institute, unpublished tabulations of DYNASIM ID914 and DYNASIM ID938 projections. 72
73
Home equity falling in CA, rising in TX Median home equity per capita, homeowners, projected to 2040 (2015 $000) $150 $125 $100 $75 $50 $25 $0 CA
TX
White NH
CA
TX
Black NH 2010
2020
CA
TX
Other NH 2030
CA
TX
Hispanic
2040
Source: Karen Smith et al., Urban Institute, unpublished tabulations of DYNASIM ID914 and DYNASIM ID938 projections. Dollars are wage-adjusted 2015 values. 73
74
Seniors’ home equity threatened in high-cost states Median per capita home equity (2015 $000) at age 75
Homeownership at age 75
$250
100% 75%
TX CA
$200 $150
50%
CA
$100 25%
TX
$50 $0
0% 2010
2020
2030
2040
2010
2020
2030
2040
Source: Karen Smith et al., Urban Institute, unpublished tabulations of DYNASIM ID914 and DYNASIM ID938 projections. Dollars are wage-adjusted 2015 values. 74
75
Policy implications: Federal, state, and local No silver bullets: we need every solution For people, we need policies that will raise incomes and wages over the life course in all states reduce income insecurity in all states
For housing, we need federal, state, and local actions to Guarantee stable affordable housing in safe neighborhoods for extremely low income people in all states For high-cost states especially: Boost housing supply by reducing regulatory burdens and investing in infrastructure Facilitate access to ownership for younger households Facilitate efficient use of homes and lots by seniors Phase out/redirect mortgage interest and property tax deductions 75
Homeownership and Household Wealth
James H. Carr Coleman A. Young Chair and Professor In Urban Affairs Wayne State University And Visiting Fellow, The Roosevelt Institute
At the Urban Institute
Annual Urban/CoreLogic Symposium Washington, DC November 2, 2016
76
1. Diverse populations 2. Diverse land use within same community 3. Walkable communities 4. Access to mass transit 5. Diverse housing stock 6. Boutique restaurants and retail/rich nightlife 7. Historic landmarks and art and cultural institutions 8. Universities and other centers of learning 9. Centralized location within metropolitan areas 10. Often bordered by impressive waterfronts
Source: Current Population Survey/Housing Vacancy Survey, 2001–14; Survey of Consumer Finances, 2001– 13); Wall Street Journal.
77
1. Diverse populations 2. Diverse land use within same community 3. Walkable communities 4. Access to mass transit 5. Diverse housing stock 6. Boutique restaurants and retail/rich nightlifeOften bordered by impressive waterfronts
Source: Author’s calculations of HMDA data, 2000–14.
78
Source: Current Population Survey/Housing Vacancy Survey, Source: eMBs, CoreLogic, HMDA, IMF, Urban Institute.
79
Source: Urban Institute calculations from HM DA and CoreLogic data. Note: Shares are computed within each race and ethnicity group. Declines are the percent decline in loans from 2001 to 2013.
80
Source: www.supercomputinginengineering.com
81
.
Source: CFED, Institute for Policy Studies; Wall Street Journal.
82
Source: The Racial Wealth Gap: Why Policy Matters. Demos and the Institute on Assets and Social Policy. Washington, DC. 2015.
83
Source: The Racial Wealth Gap: Why Policy Matters. Demos and the Institute on Assets and Social Policy. Washington, DC. 2015.
84
•
Require all federal mortgage agencies to use the most current and predictive credit scoring models on the market
•
Eliminate GSE loan level (risk-based) pricing
•
Return GSE g-fees and FHA MMPs to levels that reflect future projected losses
•
•
•
•
Leverage distressed property sales by all federal housing agencies to better promote affordable homeownership opportunities Hold private lenders accountable for exclusionary lending practices Allow the GSEs to reserve for future losses or establish a Treasury fund (within the conservatorship framework) for losses GSE losses Reform the housing finance system to address the multifaceted housing and community investment needs of America’s distressed communities into the 21st Century 85
86
Susanne V. Slater, President & CEO Habitat for Humanity of Washington, D.C.
WHY HOUSING EQUITY MATTERS 87
www.dchabitat.org
87
What Is DC’s problem? • As in other U.S. cities, market forces are driving inmigration of younger, white, and more highly-educated populations and forcing out-migration of low- to moderate-income people of color • Very limited land boundaries accelerate problem because moving further out means moving out of the city altogether
Source: “DISTRICT OF CHANGE: GENTRIFICATION AND DEMOGRAPHIC TRENDS IN WASHINGTON, DC” Chicago Policy Review, July 23, 2014 88
www.dchabitat.org
88
The Funding Gap
Current Housing Production Trust Fund Value as a Percent of Total Need
$100 Million
$5 Billion
Source: "Will D.C.’s Housing Ever Be Affordable Again?" The Atlantic, August 19, 2016 via D.C. Fiscal Policy Institute
89
www.dchabitat.org
89
Goal: Preserve Diversity & Inclusion
1. Maximize cost-effectiveness of solutions by anchoring affordable housing in neighborhoods prior to gentrification 2. Substantially increase low-income homeownership to address wealth gap 3. Allow low-income homeowners to reap expected gains 4. Balance homeownership with permanently-affordable rentals 90
www.dchabitat.org
90
What does Habitat do?
• 15th largest homebuilder in the U.S. • Provides affordable homeownership opportunities to low- to moderateincome families • Utilizes private, government, and philanthropic funding • Builds in pre-gentrified neighborhoods • Utilizes innovative, cost-effective approaches including Inclusionary Zoning projects, green building, and voluntourism 91
www.dchabitat.org
91
Case Study: Ivy City
• Worked with a coalition of nonprofits (DC Habitat, Manna Inc., Mi Casa), to build in the most-distressed census tract in the city • In a neighborhood with a population of